Artificial intelligence for construction planning and scheduling: an empirical PLS-SEM analysis of opportunities and barriers

Artificial intelligence (AI) has strong potential to enhance construction planning and scheduling through predictive analytics, generative optimization, and data-driven decision support. However, adoption in scheduling practice remains limited due to technological, organizational, and environmental constraints. This study empirically examines AI adoption in construction planning and scheduling using survey data from 204 professionals and a reflective–reflective second-order PLS-SEM model grounded in the Technology–Organization–Environment (TOE) framework and informed by Paradox Theory. Results indicate that technological and organizational dimensions are the primary drivers of both perceived barriers (β = 0.947; β = 0.905) and perceived opportunities (β = 0.953; β = 0.932). A positive relationship between overall barriers and opportunities (β = 0.151) suggests that industry challenges shape where AI is viewed as most valuable in scheduling workflows. Key opportunities; improved planning, optimal scheduling, and data-driven decision support, directly correspond to major barriers, including data quality issues, interoperability gaps, high implementation costs, and skill shortages. Interpreted through Paradox Theory, these findings show that barriers not only hinder adoption but also define the problem space where AI solutions are most needed, offering theoretical and practical guidance for prioritizing AI initiatives in scheduling.